Phased-MIMO Radar a Tradeoff Between Phased Array and MIMO Radars
Estimating Impacts of Mutual Interference of Automotive Radars · and simulation, we used a generic...
Transcript of Estimating Impacts of Mutual Interference of Automotive Radars · and simulation, we used a generic...
Estimating Impacts of Mutual Interference
of Automotive Radars
SAE Government Industry MeetingJanuary 24-26, 2018
Radar Congestion Study Purpose• This study characterizes the
environment in which automotive radars must operate
Radar Congestion Study Purpose• This study characterizes the
environment in which automotive radars must operate
• Especially as more systems with greater autonomy enter the market
• Systems that operate well in environments without other radars may suffer significant degradation of performance in radar congested environments
Why worry?
• As autonomous systems move to market, vehicles will be instrumented with multiple radars. Industry trends show increase in bandwidth and duty factor.
Figure is an example deployment of multiple radar sensors used for active safety and assisted driving systems [Kissinger , 2012]
24 GHz spectrum being phased out
Study Approach• The models for automotive radars in the
study are based on technical literature and interviews with OEMs and Tier 1 suppliers
• The modeling and simulation work focuses on two questions:– How much power does, a given
radar, receive from other radar transmitters?
– How does this impact the performance of a collision warning system?
Heuel, SteffenMicrowave Journal; Dec 2016; 59, 12; pg. 22
Estimating Interference
• The first question is answered by 𝐾𝐾−1developing a model for nominal 𝜉𝜉𝐾𝐾 = 1 −� 1 − 𝐶𝐶𝐹𝐹𝑘𝑘
𝑘𝑘=1automotive radars – Compute probability of intercept in 𝐾𝐾−1
Τ = 1 −� 1 − 𝐷𝐷𝐹𝐹spectrum, 𝜉𝜉𝐾𝐾, and time, Τ𝐾𝐾, directly 𝐾𝐾 𝑘𝑘𝑘𝑘=1
from the radar parameters, ∞ 𝑃𝑃 𝛾𝛾– Use stochastic geometric model to IK = 𝜉𝜉𝐾𝐾Τ𝐾𝐾𝜆𝜆�
0 12 2 𝑑𝑑𝑟𝑟𝐿𝐿 + 𝑟𝑟incorporate antenna pattern and 𝛿𝛿
compute interference power, IK
Estimating Impact
• The second question is answered by introducing the interference power into a system simulation. – Interference is treated as noise
(assumes waveform diversity)– Basic tracker model developed
using Matlab’s ADAS Toolbox, with reasonably chosen parameters.
Long Range Radar
Medium Range Radar
Short Range Radar
Units
Mean Transmitter Power �𝑃𝑃𝑆𝑆 1 0.3 0.1 Watts
Reference Target (Range, RCS) 100, 0(175, 10)
50, 0(88, 10)
20, 0(35, 10)
(meters, dBm2)
Transmitter Bandwidth 𝐵𝐵TX 200 400 500 MHz
Range resolution 𝑐𝑐2 𝐵𝐵𝑇𝑇𝑇𝑇
0.75 0.375 0.3 meters
Range bins 200 200 60 #
Compression Gain 23 23 18 dB
Carrier Frequency 76-7776-81
76-7776-81
76-7776-81
GHzGHz
Noise Factor, 𝑓𝑓𝑁𝑁 10 10 10 Ratio
Duty Factor DF 0.5 0.9 1 Ratio
FOV 𝜃𝜃 Azimuth 20 90 150 Degrees
FOV 𝜃𝜃 Elevation 5 10 10 Degrees
Antenna Gain 27 20 17 dB
Azimuth Resolution 5 15 50 Degrees
Range rate limits [-100 100] [-100 100] [-100 100] m/s
Generic Automotive Radar Parameters
• For the purposes of modeling and simulation, we used a generic class of radar sensor parameters
• Based on values selected from radar specifications and automotive radar experts.
• The resulting parameters are shown in table at right.
Ego Vehicle
Target Vehicle
Interferer
Scenarios in Study
Scenario 1: Opposing Traffic
For the case of opposing traffic the power from the target vehicle is computed at the reference range (10 dBsm at 100 m)
𝑃𝑃𝑇𝑇 𝑅𝑅 =𝑃𝑃𝑡𝑡𝑡𝑡𝐺𝐺𝑡𝑡𝑡𝑡𝐺𝐺𝑟𝑟𝑡𝑡𝜆𝜆2𝜎𝜎𝑇𝑇
4𝜋𝜋 3𝑅𝑅4
The results in the table consider opposing traffic, distributed as a Poisson process with average density of 1 car per 15 meters.
dB WattsEgo
RadarOther Radar
PT PC PI
76-77 GHzLRR LRR -107 -128 -69
MRR MRR -115 -138 -80
76-81 GHzLRR LRR -107 -129 -76
MRR MRR -115 -138 -87
Blue = Ego RadarGreen = Target Vehicle
Yellow = Interferers
Interference power exceeds power of target returns by more than 30 dB
Scenario 5: Rear-view Short Range Radar Illuminated by Long Range Radar
𝑃𝑃𝐼𝐼 𝑅𝑅 interference
𝑃𝑃𝑇𝑇 𝑅𝑅 target return
• The LRR has more power and greater antenna gain
• The graph shows the blue radar power reflected from the yellow car
𝑃𝑃𝑇𝑇 𝑅𝑅 =𝑃𝑃𝑡𝑡𝑡𝑡𝐺𝐺𝑡𝑡𝑡𝑡𝐺𝐺𝑟𝑟𝑡𝑡𝜆𝜆2𝜎𝜎𝑇𝑇
4𝜋𝜋 3𝑅𝑅4
• The interference from the yellow radar is calculated by Friis equation
𝑃𝑃𝐼𝐼 𝑅𝑅 =𝑃𝑃𝑇𝑇𝑇𝑇,𝐼𝐼𝐼𝐼𝑡𝑡𝐺𝐺𝐸𝐸𝐸𝐸𝐸𝐸𝐺𝐺𝐼𝐼𝐼𝐼𝑡𝑡𝜆𝜆2
4𝜋𝜋 2𝑅𝑅2
Interference power exceeds power of target returns by more than 50 dB
Scenario 5, case 2
Simulation in Matlab ADAS Toolbox
Impact on performance
• The plots show the range at which a track is formed that persists to collision• Without interference this occurs near 190 meters• With interference this occurs near 30 meters
Plot of persistent target tracks with interference (76-81 GHz)Plot of persistent target tracks from Scenario 1, long-range radar with no
interference
SINR and impact on Track Range for the Scenarios using the 76-81 GHz band
• Using generic automotive radar parameters, and tracking algorithm, terminal track ranges were reduced to as little as 16% of the range without interference
.
Not simulated
Mitigation StrategiesID
Counter Measures
Interference Reduction
Comment
MOSARIM T6.2
Detect interference and repair Rx results (time domain)
3-20 dB , depending on environment
The influence of fast or slow crossing FM chirps still needs further investigation on mitigation margin impact
RCS Study Stretch processing 10 dBThe main cost of the stretch processing technique is the loss of signal to noise ratio. So long as the interference is at least 10 dB greater than the noise, the technique is advisable.
MOSARIM T5.4
Digital Beam Forming 5-10 dBMitigation effect depends on beamwidth (space domain), based on number of elements in receiver array
MOSARIM T1.2
Specific polarization following the Radar location *
10-15 dB for co pol -systems using the same convention
This is already partially used for ACC radars that have 45 degree slant linear polarization (reduced interference from oncoming radars by 15 dB). Requires harmonization.
RCS StudySpectrum division following Radar
location *
60 to 80 dB for forward and rearward facing radars in traffic
As all automotive radars move to W-band, 76-81 GHz, splitting the spectrum could reduce interference between forward and rearward looking radars by 60 to 80 dB. Requires harmonization.
* front, side, or rear of vehicle
Conclusion• Automotive radar must perform in a challenging environment, which
will become more challenging as a greater density of advanced driver assist, and autonomous systems, populate roadways
• The Radar Congestion Study estimates these level will have significant impact on existing systems, and has tabulated the most promising mitigation strategies in practice
• Existing strategies promise to mitigate the impacts of interference and allow for good performance, but some require industry harmonization
• There is a need for methods to evaluate these strategies in automotive safety tests
ESTIMATING IMPACTS OF MUTUAL INTERFERENCE OF AUTOMOTIVE RADARS
This research is conducted by members of Michigan Tech Research Institute and University of Michigan’s Transportation Research Institute, and is supported by the National Highway Traffic Safety AdministrationDTNH22-14-D-00329L
William Buller, Michigan Technological University [email protected]